Object tracking from unmanned aerial vehicle using Kanade-Lucas-Tomasi tracker and speeded up robust features

Object detection and tracking are important and challenging tasks in many computer vision applications such as surveillance, vehicle navigation, and autonomous robot navigation. Video surveillance in a dynamic environment, especially for humans and vehicles, is one of the current challenging research topics in computer vision. In this thesis a tracking and location method for Unmanned Aerial VehicleVision System called UAV tracker is proposed. In this method, the target is extracted from the Region of Interested (ROI) automatically by Speeded Up Robust Features (SURF); then the Kanade–Lucas–Tomasi tracker is used to get the target’s position in the sequence images. The proposed framework learns about target appearance by updating the object module in each frame, which can further improve the robustness of tracker as well as feature extraction and matching process. Extensive experimental results are provided by comparing proposed algorithm with (15) related approaches on (15) challenging sequences, which demonstrate the robust tracking achieved by proposed tracker. Experimental results show that the proposed method deals with translation, rotation, partial occlusion, deformation, pose, scale changes, similar appearance and illumination change successfully